Method for calibrating relation between actual road distance and corresponding pixel distance in traffic video

Abstract

The invention provides a method for calibrating a relation between an actual road distance and a corresponding pixel distance in a traffic video. The method comprises the following steps of: dividing a road in a video image window into a plurality of sections, which are uneven in length and are connected end to end, along a road driving direction; establishing a secondary function relation between the actual road distance and the pixel distance in a corresponding video image along the driving direction in each section; calculating a secondary function coefficient of each section according to each road section end point measured in a road scene, the actual distance in one point and the corresponding pixel distance in the video image, so as to determine the secondary function relation between each section of the actual road distance and the pixel distance in the corresponding video image, and further obtain a sectioned secondary function relation between the actual road distance in the whole video window and the corresponding pixel distance. Compared with the prior art, the operation is greatly simplified, data including camera data, camera lens data, erections position and directions of a camera lens and the like do not need to be known, and the precision can be guaranteed. Compared with other linear function representations or other sectioned representation methods, the calibration precision is greatly improved.

Description

technical field [0001] This patent relates to the technical field of traffic video image signal processing, specifically, a calibration method for establishing the relationship between the actual road distance and the corresponding pixel distance in the traffic video image. Background technique [0002] With the development of transportation and information technology, people pay more and more attention to intelligent transportation. Through traffic monitoring cameras, people can observe real-time video images of roads everywhere in the control center. Due to the large and dense road network in a city, it is impossible for people to analyze the road conditions of various places through road video images one by one, in real time, artificially, and to provide quantitative traffic data such as traffic flow, average speed, and vehicle queue length. These data are automatically obtained through computer image and video processing techniques. However, the original length data pr...

AI Technical Summary

Problems solved by technology

Such as the precise focal length, the height of the camera from the ground, several azimuths of the camera, the center position of the pixel points of the camera CCD target surface, the spacing of the pixel points of the camera CCD target surface, lens distortion parameters, etc., by analyzing the ground coordinate system and the camera coordinate system , image coordinate system, etc., to establish a complex mathematical model of the imaging process, but for the application of large-scale monitoring such as traffic cameras, some data have large errors in actual operation, or even cannot be obtained at all. Therefore, the application of such methods in traffic video surveillance is greatly limited.

Images